English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Tools annual

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Tools monthly

Global: Zendy Plus monthly

Presents the access of premium content as premium feature

Premium Content

Global: Zendy Plus annual

Global: Zendy Free Plan

In this paper, we propose a method for intermediating multiple speakers'attributes and diversifying their voice characteristics in ``speakergeneration,'' an emerging task that aims to synthesize a nonexistent speaker'snaturally sounding voice. The conventional TacoSpawn-based speaker generationmethod represents the distributions of speaker embeddings by Gaussian mixturemodels (GMMs) conditioned with speaker attributes. Although this method enablesthe sampling of various speakers from the speaker-attribute-aware GMMs, it isnot yet clear whether the learned distributions can represent speakers with anintermediate attribute (i.e., mid-attribute). To this end, we propose anoptimal-transport-based method that interpolates the learned GMMs to generatenonexistent speakers with mid-attribute (e.g., gender-neutral) voices. Weempirically validate our method and evaluate the naturalness of syntheticspeech and the controllability of two speaker attributes: gender and languagefluency. The evaluation results show that our method can control the generatedspeakers' attributes by a continuous scalar value without statisticallysignificant degradation of speech naturalness.

Mid-attribute speaker generation using optimal-transport-based interpolation of Gaussian mixture models